Amplitude speeds up in-app search functionality 1000x with Algolia Search

Search slow down at scale

Collaboration is key to achieving success, says Ryan Kahn, Engineering Manager who acts as one of that team’s leads. 

Kahn’s team focuses on packaging insights into content such as dashboards and notebooks, and further driving impact to their customers by making these insights both easy to discover and distribute. Making this content fast and easy to discover is crucial for turning insight into actions. 

Algolia started working with Amplitude more than three years ago, and before then, Amplitude was using its own database tables that were not optimized for the ability to scan large volumes of data or perform matches against multiple search criteria. As Amplitude grew and added more enterprise customers this approach hit a wall: searches could take up to 30 seconds in some worst-case scenarios. 

“The biggest thing for collaboration is helping people find useful content — the things that are useful for them. Within Amplitude if you create a chart that’s useful and only you look at it, it probably wasn’t that useful, and you probably didn’t make a business decision on it.” Ryan Kahn, Engineering Manager, Amplitude

Today, Amplitude has helped nearly 1,600 companies — 26 from the Fortune 100 — innovate around more than 45,000 products and do it faster and smarter by answering one strategic question: "How do our digital products drive our business?” 

For example, Amplitude customers use its Digital Optimization System to democratize data across the organization — giving access to everyone who needs it from marketers and product managers to data scientists and executives. This means product development decisions are driven by user behavior data, and not gut calls. 

Not surprisingly, Amplitude’s own product development decisions are customer-focused and tend to come from the bottom up, with the product managers, designers, and engineers for a particular solution category — internally called pillars — working closely to solve specific customer challenges. Always keeping to the company’s three core values: humility, ownership, and growth mindset.

Improving search with flexibility and ease

Amplitude adopted Algolia’s Search across its website and in-app search functionality, with a focus on as-you-type search results and replica indexes so users can select from a sorted list. After consideration, they did not implement autocomplete in its collaboration app, since InstantSearch was delivering the results needed just as quickly. 

Implementation was smooth and extremely easy with only a few caveats, Kahn suggests: “Everything was built into Algolia, making it really easy to implement and integrate against.” 

Since Amplitude requires data encryption at rest to protect customer privacy, it selected an Enterprise Plan, and chose to run Algolia on its own servers to eliminate any usage fees. One “tricky” element, Kahn says, was integrating the company’s complex, very specific permissions model. 

Thanks to Algolia, Search infrastructure and performance has become something Kahn and his team don’t have to worry about nor require deep technical knowledge of to maintain. 

“The design of the InstantSearch React library is really elegant and helps us build the things we want and the way we want them with flexibility out of the box,” Kahn says. “In fact, things that we know we can get from Algolia are generally less work than fetching it internally or building out systems internally to fetch that data.” 

While primarily used by the collaboration team as an OEM solution in its own application and for site search, Algolia has found application across other Amplitude teams thanks to the speed at which it can surface the data teams need.

“We acknowledged early on that it was not efficient, but it was fast enough for our users. But as we started hitting scale, we realized it was not at all an OK solution, so we started researching other options.” 

Those options, Kahn says, came down to ElasticSearch or Algolia, as the only two solutions that could meet their needs. The company had already used ElasticSearch briefly but stopped due to maintenance issues and concerns around its ability to scale. 

“We asked: how do we scale this out long-term? How do we handle some edge cases? There was a little bit of performance we were trying to tune, and we decided it’s just not worth it for us, from an engineering perspective, to build that in-house knowledge that’s strong enough to make decisions when we can use Algolia out of the box.”

Powering better collaboration through faster search

Algolia Instant Search powers the search of charts, dashboards, notebooks, and cohorts and does so with incredibly fast results. Speed of search in the application increased to 30 milliseconds using Algolia, 1000x the initial speed. 

In fact, the results were so incredible they weren’t initially believed. “When I rolled this out, I actually had a lot of people at Amplitude say, ‘This is a fake demo, right?’ Because it was so fast. They’d hit the search button and the results were there before the animation happened. It was such a dramatic improvement; it just blew everyone’s mind”.

Customers' interest in using Search increased and the customer experience has improved with the same dramatic results. In the past, customers rarely used its 7 search functionality, but today Search plays a growing role in Amplitude getting customers the insights they need. 

Kahn appreciates the speed at which Algolia works — his team can resync the company’s entire production database in less than a minute and reindex it in less than 10 seconds — but also the ability to take a “set it and forget it” approach to Search, knowing security is already built-in right out of the box. 

Since its initial implementation of Algolia, Amplitude has turned to the Search provider to power a feature called Spaces, which helps teams work together faster on product development. Algolia is being used to power the capability, creating a fast, consistent, and easy-to-use experience for all users.